danielcweeks commented on PR #15045: URL: https://github.com/apache/iceberg/pull/15045#issuecomment-3751664990
> I think it would be helpful to clarify this, regardless of the decision we make. > > Personally, I agree with @gaborkaszab that adding `?snapshots=refs` and `?snapshots=all` doesn’t provide significant value. This reasoning applies to any response changes triggered solely by client requests. This is only provided in the description as an example (possibly the only known case?) to illustrate why a seemingly singular request might return different results. > Another point worth considering is time-restricted access credentials. In such cases, it might be useful for the server to update the ETag when an old credential expires or is about to expire. However, the client and server would still need a clear contract regarding the remaining validity period of credentials when an ETag is accepted. We already have the expiration time in the storage credential returned on load. If that's something that should be considered, the client implementation can bound the cache time to the credential time. I don't believe we need to spec that separately as there's already a field for that information. @gaborkaszab this might be a follow-on to make sure we aren't caching longer than access would reasonably cover, but we're currently at 5min cache and credential access is typically longer than that. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
